pith. sign in

arxiv: 2605.18893 · v2 · pith:HB5J4M6Dnew · submitted 2026-05-17 · 💻 cs.LG

Position: Graph Condensation Needs a Reset -- Move Beyond Full-dataset Training and Model-Dependence

Mridul Gupta , Samyak Jain , Vansh Ramani , Hariprasad Kodamana , Sayan Ranu This is my paper

Pith reviewed 2026-05-22 00:54 UTC · model grok-4.3

classification 💻 cs.LG
keywords graph condensationgraph neural networksgradient matchingscalabilitymodel dependenceevaluation protocolssynthetic graphsefficiency
0
0 comments X

The pith

Graph condensation methods that use gradient matching must train on the full original dataset, which defeats their efficiency purpose.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper contends that dominant graph condensation techniques, centered on gradient matching, create a built-in contradiction by needing a model trained on the complete large graph to produce the smaller synthetic version. This approach brings high computational overhead, fails to generalize across different graph neural network architectures, and depends on brittle model-specific setups. Evaluation practices in the field rely on misleading measures such as node compression ratios that ignore real resource costs and condensation time. The authors position these issues as systemic barriers and urge a shift to condensation methods that are lightweight, avoid full-dataset training, and remain independent of any particular model architecture.

Core claim

Graph condensation in its current form, dominated by gradient matching, fundamentally contradicts its own goals by necessitating full-dataset training and model dependence, leading to high overhead, poor generalization, and misleading evaluations; the field requires a reset toward lightweight and architecture-agnostic approaches.

What carries the argument

Gradient matching, the process of aligning gradients computed on the synthetic graph with those from the full original graph during model training.

Load-bearing premise

The shortcomings of gradient matching methods are systemic flaws in the current paradigm rather than issues that could be fixed by incremental improvements.

What would settle it

A successful demonstration of a graph condensation method that achieves comparable performance without any training on the full original dataset and works across multiple GNN models.

read the original abstract

Graph Neural Networks (GNNs) are powerful tools for learning from graph-structured data, but their scalability is increasingly strained by the size of real-world graphs in domains like recommender systems, fraud detection, and molecular biology. Graph condensation -- the task of generating a smaller synthetic graph that retains the performance of models trained on the original -- has emerged as a promising solution. However, the dominant approach of gradient matching introduces a fundamental contradiction: it requires training on the full dataset to create the compressed version, thereby undermining the goal of efficiency. Worse still, these methods suffer from high computational overhead, poor generalization across GNN architectures, and brittle reliance on specific model configurations. Equally concerning is the community's reliance on misleading evaluation protocols such as node compression ratios, which fail to reflect true resource savings, condensation overhead, and illusory application to neural architecture search. These shortcomings are not incidental -- they are systemic, and they obstruct meaningful progress. In this position paper, we argue that graph condensation, in its current form, needs a reset. We call for moving beyond full-dataset training and model-dependent design, and instead advocate for methods that are lightweight, architecture-agnostic, and practically deployable. By identifying key methodological flaws and outlining concrete research directions, we aim to reorient the field toward approaches that deliver on the true promise of condensation: efficient, generalizable, and usable GNN training at scale.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. This position paper argues that graph condensation requires a reset because the dominant gradient matching methods contain a fundamental contradiction: they require training on the full original dataset to produce the condensed synthetic graph, which undermines the efficiency goal. The paper further criticizes these methods for high computational overhead, poor generalization across GNN architectures, brittle model-dependence, and misleading evaluation protocols such as node compression ratios that do not capture true resource savings or applicability to neural architecture search. It calls for shifting to lightweight, architecture-agnostic, and deployable approaches.

Significance. If the identified flaws prove systemic rather than incidental, the paper could meaningfully redirect research in graph condensation toward more practical methods, supporting scalable GNN use in large domains like recommender systems and molecular biology. As a position paper it offers a clear call to action that may stimulate new directions beyond current model-dependent paradigms.

major comments (2)
  1. [Abstract] Abstract: the central claim that gradient matching 'requires training on the full dataset to create the compressed version, thereby undermining the goal of efficiency' treats the condensation step as inherently defeating efficiency. This overlooks the standard view of condensation as a one-time amortized preprocessing cost whose overhead may be offset by repeated savings in downstream training or architecture search; the position would be strengthened by quantifying or bounding this amortization.
  2. [Abstract] Abstract: the statement that the shortcomings 'are not incidental -- they are systemic, and they obstruct meaningful progress' is load-bearing for the call to reset, yet the abstract supplies no concrete derivations, comparisons, or counter-examples showing why the issues cannot be mitigated inside the gradient-matching framework. A dedicated section with such analysis is needed to support the systemic characterization.
minor comments (1)
  1. [Abstract] Abstract: the phrase 'illusory application to neural architecture search' is used without a short supporting clause or reference; adding one sentence of clarification would improve readability for readers unfamiliar with the evaluation critique.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback and the recommendation of major revision. The comments highlight opportunities to clarify the amortization perspective and to strengthen support for our characterization of the issues as systemic. We respond to each major comment below and commit to revisions that address the concerns while preserving the core position of the paper.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claim that gradient matching 'requires training on the full dataset to create the compressed version, thereby undermining the goal of efficiency' treats the condensation step as inherently defeating efficiency. This overlooks the standard view of condensation as a one-time amortized preprocessing cost whose overhead may be offset by repeated savings in downstream training or architecture search; the position would be strengthened by quantifying or bounding this amortization.

    Authors: We agree that framing condensation purely as a one-time cost is a common perspective and that explicit discussion of amortization would strengthen the argument. However, our position is that for many real-world graphs the one-time full-dataset training cost is already prohibitive, and reported overheads in gradient-matching methods frequently exceed downstream savings even after multiple uses. In the revision we will add a quantitative discussion of amortization, including simple bounds derived from published overhead figures and scenarios (e.g., single downstream training versus repeated NAS queries) under which the net benefit materializes. revision: yes

  2. Referee: [Abstract] Abstract: the statement that the shortcomings 'are not incidental -- they are systemic, and they obstruct meaningful progress' is load-bearing for the call to reset, yet the abstract supplies no concrete derivations, comparisons, or counter-examples showing why the issues cannot be mitigated inside the gradient-matching framework. A dedicated section with such analysis is needed to support the systemic characterization.

    Authors: We accept that the abstract's strong wording requires more explicit grounding. The manuscript body already presents concrete examples of model dependence, overhead scaling, and evaluation mismatches that are difficult to resolve within gradient matching without abandoning its core mechanics. To make this support immediately visible, we will insert a dedicated subsection (likely in the introduction or a new “Why the Issues Are Systemic” section) that consolidates the key derivations, side-by-side comparisons with non-gradient-matching baselines, and counter-examples illustrating why incremental fixes inside the existing paradigm fall short. revision: yes

Circularity Check

0 steps flagged

Position paper critique relies on external descriptions of existing methods with no self-referential derivation

full rationale

The paper is a position paper whose central claims critique dominant graph condensation techniques (e.g., gradient matching requiring full-dataset training) by describing their documented properties in prior literature. No equations, fitted parameters, or predictions are introduced that reduce to the paper's own inputs by construction. Claims rest on characterizations of external methods rather than self-citations or ansatzes that would create circularity. The argument is therefore self-contained against external benchmarks and exhibits no load-bearing self-referential steps.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central argument rests on domain assumptions about what efficiency and generalizability mean in graph condensation, without introducing new free parameters or invented entities.

axioms (2)
  • domain assumption Graph condensation methods should not require training on the full original dataset to achieve efficiency gains.
    This premise underpins the claimed contradiction in gradient matching and is invoked throughout the abstract as the core flaw.
  • domain assumption Evaluation protocols based on node compression ratios fail to capture true resource savings and practical usability.
    Stated explicitly when describing misleading metrics that obstruct progress.

pith-pipeline@v0.9.0 · 5803 in / 1316 out tokens · 48382 ms · 2026-05-22T00:54:14.433112+00:00 · methodology

Review history (2 revisions) →

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

121 extracted references · 121 canonical work pages · 3 internal anchors

  1. [1]

    Proceedings of the web conference 2020 , pages=

    Graphgen: A scalable approach to domain-agnostic labeled graph generation , author=. Proceedings of the web conference 2020 , pages=

    work page 2020

  2. [2]

    IJCAI , year=

    Graphreach: Position-aware graph neural network using reachability estimations , author=. IJCAI , year=

    work page

  3. [3]

    Transactions on Machine Learning Research , issn=

    Training Graph Neural Networks Subject to a Tight Lipschitz Constraint , author=. Transactions on Machine Learning Research , issn=. 2024 , url=

    work page 2024

  4. [4]

    Advances in Neural Information Processing Systems , volume=

    Neuromlr: Robust & reliable route recommendation on road networks , author=. Advances in Neural Information Processing Systems , volume=

    work page

  5. [5]

    Advances in Neural Information Processing Systems , volume=

    Learning articulated rigid body dynamics with lagrangian graph neural network , author=. Advances in Neural Information Processing Systems , volume=

    work page

  6. [6]

    International Conference on Machine Learning , pages=

    Stridernet: A graph reinforcement learning approach to optimize atomic structures on rough energy landscapes , author=. International Conference on Machine Learning , pages=. 2023 , organization=

    work page 2023

  7. [7]

    International Conference on Machine Learning , pages=

    Grafenne: learning on graphs with heterogeneous and dynamic feature sets , author=. International Conference on Machine Learning , pages=. 2023 , organization=

    work page 2023

  8. [8]

    Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining , pages=

    Frigate: Frugal spatio-temporal forecasting on road networks , author=. Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining , pages=

    work page

  9. [9]

    arXiv preprint arXiv:2402.12937 , year=

    Graphgini: Fostering individual and group fairness in graph neural networks , author=. arXiv preprint arXiv:2402.12937 , year=

    work page arXiv

  10. [10]

    The Eleventh International Conference on Learning Representations , year=

    Enhancing the inductive biases of graph neural ode for modeling physical systems , author=. The Eleventh International Conference on Learning Representations , year=

    work page

  11. [11]

    The Twelfth International Conference on Learning Representations , year=

    BroGNet: Momentum-Conserving Graph Neural Stochastic Differential Equation for Learning Brownian Dynamics , author=. The Twelfth International Conference on Learning Representations , year=

    work page

  12. [12]

    Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V

    Persona identification in e-commerce with scarce labels and in-context graph learning , author=. Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V. 2 , pages=

    work page

  13. [13]

    ICLR , year=

    Graph attention networks , author=. ICLR , year=

    work page

  14. [14]

    Drug discovery today , volume=

    Graph neural networks for automated de novo drug design , author=. Drug discovery today , volume=. 2021 , publisher=

    work page 2021

  15. [15]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Dataset distillation by matching training trajectories , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    work page

  16. [16]

    Advanced Materials , volume=

    Machine-learning-assisted determination of the global zero-temperature phase diagram of materials , author=. Advanced Materials , volume=. 2023 , publisher=

    work page 2023

  17. [17]

    Nature Machine Intelligence , author=

    CHGNet as a pretrained universal neural network potential for charge-informed atomistic modelling , DOI=. Nature Machine Intelligence , author=. 2023 , pages=

    work page 2023

  18. [18]

    Open Materials 2024 (OMat24) Inorganic Materials Dataset and Models

    Open materials 2024 (omat24) inorganic materials dataset and models , author=. arXiv preprint arXiv:2410.12771 , year=

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  19. [19]

    ICLR , year=

    Semi-supervised classification with graph convolutional networks , author=. ICLR , year=

    work page

  20. [20]

    arXiv preprint arXiv:2402.02000 , year=

    A survey on graph condensation , author=. arXiv preprint arXiv:2402.02000 , year=

    work page arXiv

  21. [21]

    IJCAI , year=

    A comprehensive survey on graph reduction: Sparsification, coarsening, and condensation , author=. IJCAI , year=

    work page

  22. [22]

    IEEE Transactions on Knowledge and Data Engineering , year=

    Graph condensation: A survey , author=. IEEE Transactions on Knowledge and Data Engineering , year=

    work page

  23. [23]

    Sun, Qingyun and Chen, Ziying and Yang, Beining and Ji, Cheng and Fu, Xingcheng and Zhou, Sheng and Peng, Hao and Li, Jianxin and Yu, Philip S , journal=

    work page

  24. [24]

    IEEE transactions on neural networks and learning systems , volume=

    A comprehensive survey on graph neural networks , author=. IEEE transactions on neural networks and learning systems , volume=. 2020 , publisher=

    work page 2020

  25. [25]

    Advances in neural information processing systems , volume=

    Inductive representation learning on large graphs , author=. Advances in neural information processing systems , volume=

    work page

  26. [26]

    Relational inductive biases, deep learning, and graph networks

    Relational inductive biases, deep learning, and graph networks , author=. arXiv preprint arXiv:1806.01261 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  27. [27]

    2nd Workshop on Advancing Neural Network Training: Computational Efficiency, Scalability, and Resource Optimization (WANT@ ICML 2024) , year=

    Model-Agnostic Graph Dataset Compression with the Tree Mover’s Distance , author=. 2nd Workshop on Advancing Neural Network Training: Computational Efficiency, Scalability, and Resource Optimization (WANT@ ICML 2024) , year=

    work page 2024

  28. [28]

    Proceedings of the ACM on Web Conference 2024 , pages=

    Fast graph condensation with structure-based neural tangent kernel , author=. Proceedings of the ACM on Web Conference 2024 , pages=

    work page 2024

  29. [29]

    AI open , volume=

    Graph neural networks: A review of methods and applications , author=. AI open , volume=. 2020 , publisher=

    work page 2020

  30. [30]

    Workload prediction in edge computing based on graph neural network , author=. 2021 IEEE Intl Conf on Parallel & Distributed Processing with Applications, Big Data & Cloud Computing, Sustainable Computing & Communications, Social Computing & Networking (ISPA/BDCloud/SocialCom/SustainCom) , pages=. 2021 , organization=

    work page 2021

  31. [31]

    NeurIPS , year=

    Ogb-lsc: A large-scale challenge for machine learning on graphs , author=. NeurIPS , year=

    work page

  32. [32]

    ACM Computing Surveys , volume=

    Graph neural networks in recommender systems: a survey , author=. ACM Computing Surveys , volume=. 2022 , publisher=

    work page 2022

  33. [33]

    A survey on spectral graph neural networks

    A survey on spectral graph neural networks , author=. arXiv preprint arXiv:2302.05631 , year=

    work page arXiv

  34. [34]

    IEEE Transactions on Knowledge and Data Engineering , volume=

    Influence maximization on social graphs: A survey , author=. IEEE Transactions on Knowledge and Data Engineering , volume=. 2018 , publisher=

    work page 2018

  35. [35]

    arXiv preprint arXiv:2207.04869 , year=

    Graph-based molecular representation learning , author=. arXiv preprint arXiv:2207.04869 , year=

    work page arXiv

  36. [36]

    IEEE Transactions on Intelligent Transportation Systems , volume=

    Graph neural networks for intelligent transportation systems: A survey , author=. IEEE Transactions on Intelligent Transportation Systems , volume=. 2023 , publisher=

    work page 2023

  37. [37]

    ACM Transactions on Information Systems , volume=

    Automl for deep recommender systems: A survey , author=. ACM Transactions on Information Systems , volume=. 2023 , publisher=

    work page 2023

  38. [38]

    NeurIPS 2022 Workshop: New Frontiers in Graph Learning , year=

    Faster hyperparameter search on graphs via calibrated dataset condensation , author=. NeurIPS 2022 Workshop: New Frontiers in Graph Learning , year=

    work page 2022

  39. [39]

    Pattern Recognition , pages=

    Continual graph learning: A survey , author=. Pattern Recognition , pages=. 2026 , publisher=

    work page 2026

  40. [40]

    Tsinghua Science and Technology , volume=

    Graph neural architecture search: A survey , author=. Tsinghua Science and Technology , volume=. 2021 , publisher=

    work page 2021

  41. [41]

    International conference on machine learning , pages=

    Neural message passing for quantum chemistry , author=. International conference on machine learning , pages=. 2017 , organization=

    work page 2017

  42. [42]

    ICLR , year=

    Fastgcn: fast learning with graph convolutional networks via importance sampling , author=. ICLR , year=

    work page

  43. [43]

    ICLR , year=

    Graphsaint: Graph sampling based inductive learning method , author=. ICLR , year=

    work page

  44. [44]

    International conference on machine learning , pages=

    Simplifying graph convolutional networks , author=. International conference on machine learning , pages=. 2019 , organization=

    work page 2019

  45. [45]

    Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining , pages=

    Graph data condensation via self-expressive graph structure reconstruction , author=. Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining , pages=

    work page

  46. [46]

    International Conference on Machine Learning , pages=

    NAFS: a simple yet tough-to-beat baseline for graph representation learning , author=. International Conference on Machine Learning , pages=. 2022 , organization=

    work page 2022

  47. [47]

    IEEE Transactions on Knowledge and Data Engineering , year=

    Acceleration algorithms in gnns: A survey , author=. IEEE Transactions on Knowledge and Data Engineering , year=

    work page

  48. [48]

    IEEE Transactions on Big Data , volume=

    Data-centric graph learning: A survey , author=. IEEE Transactions on Big Data , volume=. 2024 , publisher=

    work page 2024

  49. [49]

    arXiv preprint arXiv:1811.10959 , year=

    Dataset distillation , author=. arXiv preprint arXiv:1811.10959 , year=

    work page arXiv

  50. [50]

    International Conference on Learning Representations , year=

    Dataset Condensation with Gradient Matching , author=. International Conference on Learning Representations , year=

    work page

  51. [51]

    Proceedings of the fortieth annual ACM symposium on Theory of computing , pages=

    Graph sparsification by effective resistances , author=. Proceedings of the fortieth annual ACM symposium on Theory of computing , pages=

    work page

  52. [52]

    International conference on machine learning , pages=

    Spectrally approximating large graphs with smaller graphs , author=. International conference on machine learning , pages=. 2018 , organization=

    work page 2018

  53. [53]

    Advances in Neural Information Processing Systems , volume=

    A unifying framework for spectrum-preserving graph sparsification and coarsening , author=. Advances in Neural Information Processing Systems , volume=

    work page

  54. [54]

    Aditya and Jin, Wei , title =

    Hashemi, Mohammad and Gong, Shengbo and Ni, Juntong and Fan, Wenqi and Prakash, B. Aditya and Jin, Wei , title =. 2024 , isbn =. doi:10.24963/ijcai.2024/891 , booktitle =

    work page doi:10.24963/ijcai.2024/891 2024

  55. [55]

    ICLR , year=

    Graph condensation for graph neural networks , author=. ICLR , year=

    work page

  56. [56]

    Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining , pages=

    Condensing graphs via one-step gradient matching , author=. Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining , pages=

    work page

  57. [57]

    Advances in Neural Information Processing Systems , volume=

    Does graph distillation see like vision dataset counterpart? , author=. Advances in Neural Information Processing Systems , volume=

    work page

  58. [58]

    Advances in Neural Information Processing Systems , volume=

    Fair graph distillation , author=. Advances in Neural Information Processing Systems , volume=. 2023 , url=

    work page 2023

  59. [59]

    arXiv preprint arXiv:2402.04924 , year=

    Two trades is not baffled: Condensing graph via crafting rational gradient matching , author=. arXiv preprint arXiv:2402.04924 , year=

    work page arXiv

  60. [60]

    arXiv preprint arXiv:2311.15772 , year=

    Attend who is weak: Enhancing graph condensation via cross-free adversarial training , author=. arXiv preprint arXiv:2311.15772 , year=

    work page arXiv

  61. [61]

    Proceedings of the ACM on Web Conference 2024 , pages=

    Exgc: Bridging efficiency and explainability in graph condensation , author=. Proceedings of the ACM on Web Conference 2024 , pages=

    work page 2024

  62. [62]

    International Conference on Learning Representations , year=

    How Powerful are Graph Neural Networks? , author=. International Conference on Learning Representations , year=

    work page

  63. [63]

    Qiying Pan and Ruofan Wu and Tengfei Liu and Tianyi Zhang and Yifei Zhu and Weiqiang Wang , booktitle=. Fed. 2023 , url=

    work page 2023

  64. [64]

    2024 IEEE 40th International Conference on Data Engineering (ICDE) , pages=

    Graph condensation for inductive node representation learning , author=. 2024 IEEE 40th International Conference on Data Engineering (ICDE) , pages=. 2024 , organization=

    work page 2024

  65. [65]

    arXiv preprint arXiv:2206.13697 , year=

    Graph condensation via receptive field distribution matching , author=. arXiv preprint arXiv:2206.13697 , year=

    work page arXiv

  66. [66]

    IEEE Transactions on Knowledge and Data Engineering , volume=

    Puma: Efficient continual graph learning for node classification with graph condensation , author=. IEEE Transactions on Knowledge and Data Engineering , volume=. 2024 , publisher=

    work page 2024

  67. [67]

    2023 IEEE International Conference on Data Mining (ICDM) , pages=

    Cat: Balanced continual graph learning with graph condensation , author=. 2023 IEEE International Conference on Data Mining (ICDM) , pages=. 2023 , organization=

    work page 2023

  68. [68]

    Proceedings of the ACM on Web Conference 2025 , pages=

    Disentangled condensation for large-scale graphs , author=. Proceedings of the ACM on Web Conference 2025 , pages=

    work page 2025

  69. [69]

    Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining , pages=

    Kernel ridge regression-based graph dataset distillation , author=. Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining , pages=

    work page

  70. [70]

    Forty-first International Conference on Machine Learning , year=

    Graph Distillation with Eigenbasis Matching , author=. Forty-first International Conference on Machine Learning , year=

    work page

  71. [71]

    The Twelfth International Conference on Learning Representations , year=

    Mirage: Model-agnostic Graph Distillation for Graph Classification , author=. The Twelfth International Conference on Learning Representations , year=

    work page

  72. [72]

    Advances in Neural Information Processing Systems , volume=

    Structure-free graph condensation: From large-scale graphs to condensed graph-free data , author=. Advances in Neural Information Processing Systems , volume=

    work page

  73. [73]

    European Conference on Computer Vision , pages=

    GSTAM: Efficient Graph Distillation with Structural Attention-Matching , author=. European Conference on Computer Vision , pages=. 2024 , organization=

    work page 2024

  74. [74]

    Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining , pages=

    Self-Supervised Learning for Graph Dataset Condensation , author=. Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining , pages=

    work page

  75. [75]

    ICML , year=

    Navigating complexity: Toward lossless graph condensation via expanding window matching , author=. ICML , year=

    work page

  76. [76]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Bi-directional multi-scale graph dataset condensation via information bottleneck , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    work page

  77. [77]

    IEEE Transactions on Knowledge and Data Engineering , year=

    Efficient Graph Condensation via Gaussian Process , author=. IEEE Transactions on Knowledge and Data Engineering , year=

    work page

  78. [78]

    Knowledge-Based Systems , volume=

    Multiple sparse graphs condensation , author=. Knowledge-Based Systems , volume=. 2023 , publisher=

    work page 2023

  79. [79]

    The Thirteenth International Conference on Learning Representations , year=

    Bonsai: Gradient-free Graph Condensation for Node Classification , author=. The Thirteenth International Conference on Learning Representations , year=

    work page

  80. [80]

    Applied Sciences , VOLUME =

    Mao, Runze and Fan, Wenqi and Li, Qing , TITLE =. Applied Sciences , VOLUME =. 2023 , NUMBER =

    work page 2023

Showing first 80 references.